rlmmart-09s01n1-p223
TRANSCRIPT
-
7/26/2019 RLMMArt-09S01N1-p223
1/14
Suplemento de la Revista Latinoamericana de Metalurgia y Materiales2009; S1(1): 223-234
0255-6952 2009 Universidad Simn Bolvar (Venezuela) 221
DESIGN OF A DIE FOR THE COLD COMPACTION CALIBRATION OF POWDERED
MATERIALS
Mauricio Barrera*, Hctor Snchez S.
Este artculo forma parte del Volumen Suplemento S1 de la Revista Latinoamericana de Metalurgia y Materiales(RLMM). Los suplementos de la RLMM son nmeros especiales de la revista dedicados a publicar memorias decongresos.
Este suplemento constituye las memorias del congreso X Iberoamericano de Metalurgia y Materiales (XIBEROMET) celebrado en Cartagena, Colombia, del 13 al 17 de Octubre de 2008.
La seleccin y arbitraje de los trabajos que aparecen en este suplemento fue responsabilidad del ComitOrganizador del X IBEROMET, quien nombr una comisin ad-hoc para este fin (vase editorial de estesuplemento).
La RLMMno someti estos artculos al proceso regular de arbitraje que utiliza la revista para los nmeros regularesde la misma.
Se recomend el uso de las Instrucciones para Autores establecidas por la RLMM para la elaboracin de losartculos. No obstante, la revisin principal del formato de los artculos que aparecen en este suplemento fueresponsabilidad del Comit Organizador del X IBEROMET.
-
7/26/2019 RLMMArt-09S01N1-p223
2/14
-
7/26/2019 RLMMArt-09S01N1-p223
3/14
Suplemento de la Revista Latinoamericana de Metalurgia y Materiales2009; S1(1): 223-234
0255-6952 2009 Universidad Simn Bolvar (Venezuela) 223
DESIGN OF A DIE FOR THE COLD COMPACTION CALIBRATION OF POWDERED
MATERIALS
Mauricio Barrera*, Hctor Snchez S.
Escuela de Ingeniera de Materiales, Universidad del Valle. Cali, Colombia
E-mail: [email protected]
Trabajos presentados en el X CONGRESO IBEROAMERICANO DE METALURGIA Y MATERIALESIBEROMET
Cartagena de Indias (Colombia), 13 al 17 de Octubre de 2008
Seleccin de trabajos a cargo de los organizadores del evento
Publicado On-Line el 20-Jul-2009Disponible en: www.polimeros.labb.usb.ve/RLMM/home.html
Abstract
This paper addresses the problem of finding the radial stress within a powder body undergoing plastic deformation by
confined compression, motivated by the compaction test for powdered materials. Small deformation, elastoplastic
behaviour is assumed, with porosity as the parameter governing the evolution of material properties. An elliptic yieldsurface is selected, whose aspect ratio approaches zero as full density is neared, so as to recover the Von-Mises criterion of
plastic yield in ductile metals. The results were used to track the radial stresses on a notional compaction process, and
applied to the design of an instrumented closed die. In addition, the aspect ratio of the yield surface showed an evolution
according to the widely used Heckel porosity-pressure function [6], thus extending the interpretation of this model to amore general stress state.
Keywords:Powder compaction, Powder plasticity, Plasticity model calibration
Resumen
Este articulo trata el problema de hallar el esfuerzo radial dentro de un espcimen de material en polvo sometido a
deformacin plstica mediante compresin confinada, motivado por el ensayo de compactacin para materiales en aquel
estado. Se asume un comportamiento elstico-plstico de pequeas deformaciones, siendo la porosidad el parmetro quegobierna la evolucin de las propiedades del material. Se escoge una superficie elptica de fluencia plstica, cuya relacin
de aspecto se aproxima a cero en la medida que el material se acerca a la densificacin total. De este modo se recupera elcriterio de fluencia plstica de Von-Mises. Los resultados del anlisis se usaron para calcular los esfuerzos radiales duranteun proceso de compactacin propuesto para el caso, y se aplicaron al diseo de un dado de compactacin instrumentado.
Adicionalmente, la relacin de aspecto de la superficie de fluencia mostr una evolucin de acuerdo con el modelo
porosidad-presin de Heckel [6], extendiendo as la interpretacin que se hace de este modelo a una situacin de estado deesfuerzo mas general.
Palabras clave:Compactacin de polvos, Plasticidad de polvos, Calibracin de modelos de plasticidad
1 INTRODUCTION
The current context within which manufacturing
processes of powder materials are engineered
demands a mechanical characterization of thematerial, considering the multiaxial nature of thestress-strain state. To this purpose the conventional
close compaction test [1] must allow for computing
the radial stress into the specimen, in addition to the
punching -or axial- stress. In particular, a set of
strain gauges measuring the hoop strain at the outer
surface of the die is laid out to fulfil the requirement
[2,3]. Having recourse to thick walled vessel
formulae [4] leads to a prompt calculation of the
radial stress present in the powder specimen. A
crucial point is the determination of an appropriate
value for die wall thickness: too thin and it would
collapse upon loading of the powder specimen; too
thick and hoop strains could not be properly gauged.
In the light of this, it is clear that some estimation of
radial stress exerted by the powder body, which is
undergoing uniaxial confined compression, is to be
done in advance. In a first approximation to this
problem the analysis of seemingly influential
factors, such as die-wall friction and
nonuniformities within the specimen thereof -both
in axial and radial directions-, is relinquished on
-
7/26/2019 RLMMArt-09S01N1-p223
4/14
Barrera et al.
224 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
behalf of mathematical tractability. That said, it was
decided to turn to a yield surface expression devised
for aluminium foams [5], deeming it as makeshift
powdered material. The end result is a solution to
the problem of uniaxial confined compression of a
specimen of increasing relative density, where
elastic stresses are stored up until plastic onset,dictated by the mentioned material model, occurs.
In this regards, and to completely define the
problem an elastoplastic behaviour assumption
comes necessary, whereby elastic behaviour is
assumed to be linear and of small strains, and also
showing a dependence on the degree of compaction.
Elastic computations provide the stress state by
means of which plastic yielding is arrived to; the
yield function and an assumed associative plastic
behaviour provide an update of plastic straining and,
hence, of porosity. This is then used to compute new
values of elastic constants. The application of thisprocedure throughout a compaction process
permitted to track the evolution of radial stresses,
and therefore, the application of thick wall vessel
formulae to make a decision in regards of die wall
thickness. The compaction die was built and the
requirements of structural strength and hoop strain
measurement were satisfactorily met. The evolution
of the yield surface parameter, called aspect ratio,
showed to obey a Heckel-like function [6], thus
providing an interpretation of this porosity-pressure
model in terms of a more general stress state.
2 THE YIELD FUNCTION
The classic Von-Mises yield function has been
modified by several authors intending to represent
the observed behaviour in porous materials, where a
ductile, metal-like mechanical deformation at full
density is attained eventually. One such model is
given by Eq.(1):
( ) 00 = kg ij (1)
The first term represents a function of stress
components which is the equivalent stress fromVon-Mises theory, including a modification by
hydrostatic pressure. The second term stands for
yield strength in uniaxial test, whether tension or
compression. It shows a dependency on the extent of
plastic deformation through an observable
parameterk. The function of stresses is given byEq.(2) [7]:
( )( )
+
+=
2
222
31
meijg (2)
It is worth to note how this function is actually a
quadratic expression in{ }me , depicting an ellipseon ( )qp, plane [9], which has proven adequate infitting experimental data to results of triaxial test
probing of metal foams and some granular
materials. It makes part of a more general family of
elliptic yield functions of the form [8]:
0' 202
12 =+ JJ (3)
posed in terms of stress invariants ( )', 21 JJ forstress tensor and stress deviator, and two empirical,
material dependent parameters ( ), .The stress components shown in the numerator of
Eq.(2) stand for the square root of the double
contraction of stress deviator equivalent or
effective stress- and one third of the trace of
hydrostatic component of stress tensor-mean stress-,
as set out in Eq.(4) and Eq.(5):
21
2
3
= ijije SS
(4)
kkm 31=
(5)
The parameter captures the dependency of yieldfunction on hydrostatic stress. At full density,
0= and the yield function of Von-Mises isrecovered. In deriving the yield function, intended to
characterize aluminium foams, a material specimen
was probed using a triaxial load cell, and the results
of Figure 1 were obtained.
It is shown that the yielding of the material is
described by an ellipse lying on the ( )qp, planecharacterized by the ratio of its semi axes. As thisis a material-dependent quantity, the parameter isa material property called pressure-sensitivity
coefficient.
-
7/26/2019 RLMMArt-09S01N1-p223
5/14
Design of a die for the cold compaction calibration of powdered materials
Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234 225
Figure 1. Experimental yield loci for some typical
aluminium foams and comparison with elliptic yield
surface [7].
As to the coordinate axes in Figure 1, Figure 2recalls the stress state for a material element
undergoing closed die compaction, in terms of
punching pressure and radial pressure. Stress
components homogeneity throughout the specimen
is assumed.
axial
Vaxial =
Hradial =
Figure 2. Stress state of a homogeneous specimen
undergoing confined compression.
Certainly a relationship between equivalent and
mean stress with p and q components of stress may
be established, so that Eq.(6) and Eq.(7) hold [9]:
( )HVe q == (6)
3
2 HVm p
+== (7)
3 CONFINED COMPRESSION OF A
METAL FOAM
The aim is to know what the radial pressure is that a
specimen of metal foam, undergoing closed die
compaction, would exert. This material is
considered, in the approximation presented here, as
appropriately taking the place of a metal powder
body. Replacing Eq.(2) in Eq.(1) and from the
considerations leading to Eq.(6) and Eq.(7) it is
possible to state Eq.(8)
( )( )( )
( )
( )
+
+
=HVm
HVek
,
,
31
122
2
20
(8)
The observable material parameter kto the left handside of the equation has been made dependent on
density, i.e. density is the quantity used to track the
evolution of plastic deformation of the material.
As the punching stress, represented by V , is the
experimental input in a closed die compaction test,
and also as this happens to be the yield strength ofthe tested material throughout compaction, the aim
is to solve Eq.(8) for it. Nonetheless an issue is yet
to be addressed, namely, the way the radial stress
exerted against the die wall is generated. At thispoint it is assumed that elastic behaviour is
responsible for storing that energy which carries
over to exerted radial pressure. Linear elastic, small
strains behaviour is assumed at this point. This in
conjunction with presumed homogeneity within thespecimen permits to state the generalized Hookes
law to an axisymmetric element, as shown in Figure
3 and rearranged in Eq.(9).
( )[
( )[ ]
( )[ ]
+=
+=
+=
PE
PE
PE
rr
rr
rrzz
10
10
1
zz
rr
rr
zz
rz
zr
Figure 3. Axisymmetric stress state and generalized
Hooke law to the case.
=
B
A
A
P
AA
BA
AB
zz
rr
1
0
0
(9)
In Eq.(9)1 EA and 1EB . The linear
dependence of the first two equations in Eq.(9)
-
7/26/2019 RLMMArt-09S01N1-p223
6/14
Barrera et al.
226 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
shows that radial and transverse stress components
equal each other. This is why a single component is
given to them in the context of stress states on
( )qp, plane, where axisymmetric stress states areassumed (i.e. Hrr == ).
Solving for radial stress yields Eq.(10):
PKAB
APrr =
= (10)
The last term to the right is just shorthand for
upcoming derivations. This result may now be
replaced back into Eq.(8) and solved for punching
stress PVzz == , resulting in Eq.(11) below:
( )
( )( ) ( )
21
222
2
0
219
19
1
++
+
=KK
P
(11)
Eq.(11) can be thought of as the yield strength of the
material under closed die compaction. It is worth torecall that a yielding point is arrived to by means of
storing energy by previous deformation, and it is the
elastic regime of the material where this storage
takes place.
So, here lies the central assumption to the derivation
just conducted: the radial stress component is taken
to be an immediate consequence of the confined
compaction carried out by the punching stress and a
presumed linear elastic behaviour of the material. It
is this link between stress components, stemming
from the axisymmetric, confined compression
mechanical condition what allowed for the
expression in Eq.(11) to be gained.
4 EVOLUTION EQUATIONS: YIELD
SURFACE
It is left now to specify the way in which the yield
surface aspect ratio , yield strength under uniaxial
loading ( )( ) k0 and elastic properties Kdependupon a measure of plastic deformation. Relativedensity will serve the purpose, in a first
approximation since, in so doing, the second
invariant of the strain tensor is being neglected as a
measure of deformation. Then ( ) and( ) ( ){ } ,E shall be sought in the sequel.
As to the yield surface aspect ratio, it is necessary to
recall that Eq.(2) was originally intended to
represent the yield behaviour of a metal foam. Such
a material is not purported to deform plastically by
reducing its porosity up until attaining full density,
unlike metal powders during compaction. The yield
function in Eq.(2), within the context of metal
foams, is said to evolve in a self-similar manner, so
that the aspect ratio is virtually a constantthroughout the plastic deformation process. The
hardening rules to that particular case are derived
from application of Prandtl-Reuss plastic flow rule
and definition of a work conjugate stress-strain pair,
in such a way that the tangent modulus from readily
conducted tests, such as hydrostatic and uniaxial
loading, may be used to characterize the way a
material hardens plastically.
To make the case for powder compaction one could
resort to some results of triaxial probing over metal
powder specimens showing that, at least during the
first stages of compaction, the yield surface is
appropriately represented by an ellipse. Now, as the
powder is intended to attain full density, aspect ratio
must be made variable with relative density. So, an
evolution like that set out in Figure 4 is pursued.
q
p02P01P
02Q
01Q
2
1
iQ0 i
iP0
fullsurfaceyieldMisesVon
0
0
P
Q
Figure 4. Intended yield surface and its evolution with
relative density.
A hint to the functional form of ( ) is provided bythe measured evolution of the Cam-Clay yield
function as applied to some metal powders of
industrial interest. The evolution of the ellipses,
characterized by their semi axis lengths ( )00 ,QP isdescribed by the function shown in Eq.(12) [2].
=
max
01max0 tanh
Q
PKQQ (12)
It is clearly a non trivial task to solve analytically
for1
00
= PQ . Yet, from the same reference, 0P istaken as the yield surface parameter uniquely
-
7/26/2019 RLMMArt-09S01N1-p223
7/14
Design of a die for the cold compaction calibration of powdered materials
Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234 227
defining yield surface shape and also uniquely
depending on relative density. A particular form of
this dependency was found to be given by Eq.(13)
3
0
20
1ln
K
full
KP
=
(13)
In this equation ( )max0 ,, stand for currentdensity, apparent density and maximum (theoretical)
density of the powdered material, respectively.
Equations Eq(12) and Eq.(13) turned out to be
satisfactory fitting functions and are not
prescriptive, i.e. experimental data from other
materials may quite well be fitted by some other
functions. Nonetheless, it is always preferred to use
functions which are somehow related with powder
compaction models, particularly in regards ofEq.(13). Being this the case, a physical meaning of
the fitting parameters is gained. In this sense it is
worth to mention that compaction models relating
0P with , such as that of Heckel [6], are widely
employed for this specific purpose.
The case was made for an iron alloy powder, which
bore a yield strength of 170MPa ( maxQ ) after
compaction. Fitting parameters were 1.5, 21.903 and
1.3527 for { }321 ,, KKK respectively [2]. Tabulatingdata from Eq.(12) and Eq.(13) provides a mean to
track the progress ( ) sought for.
Table 1 displays the results of a compaction process
of up to 87.5% of full density for an iron powder
(density 7.8Kg/m3), which is what in practice is
reached to such a material.
Table 1.Data from a compaction process, iron powder.
Current
density
(Kg/m3)
P0(MPa) Q0(MPa)Aspect
Ratio
2.808 0.000 0.000 -
5.803 19.460 28.907 1.485
6.802 41.694 59.864 1.436
7.301 67.682 90.958 1.344
Apparent density (Kg/m3): 2.808; Theoretical
density (Kg/m3): 7.8; Qmax(MPa): 170 [2]
It can be seen that the aspect ratio changes are
negligible and that a Von-Mises yield surface might
not be recovered this way. However, assuming that
from this compaction state up to full density there
are not changes in material phenomena underlying
fitting parameters{ }321 ,, KKK , one may actuallyapproach the theoretical density (7.8 Kgm-3). Resultsare set out in Table 2 below.
Table 2.Yield surface aspect ratio on using Eq.(13) for
extrapolation.
Current
density
(Kg/m3)
P0(MPa) Q0(MPa)Aspect
Ratio
7.650 119.603 133.257 1.114
7.794 292.428 168.060 0.575
7.797 337.976 169.129 0.500
7.799 429.565 169.827 0.395
It can be seen that the yield surface actually tends to
approach Von-Mises yield surface in a noticeablemanner only as density is within 2% before
theoretical density. Then, one may take these just
tabulated results to perform a curve fitting leading to
an evolution equation ( ) . The following variablemodification, Eq.(14), is enforced so that curve
fitting results do not resort to an unwieldy set of
empirical constants, which would make a physical
interpretation nearly impossible.
( )
=
full
LnLn
1
1 (14)
Plotting this equation leads to Figure 5 below.
-1.00
-0.80
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00 9.00 10.00
( )Ln
( )
full
Ln1
1
Figure 5.Evolution of yield surface aspect ratio with a
logarithmic function of relative density.
-
7/26/2019 RLMMArt-09S01N1-p223
8/14
Barrera et al.
228 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
From the value 2.00 (corresponding to a relative
density of about 87%) for the abscissa towards the
right, a straight line would provide a satisfactory
fitting, indicating an exponential-like relationship
between the plotted parameters. So, the
dependence ( ) could be split into two ranges:
From apparent density to 87% full density: aspect
ratio is almost a constant, with a value of 5.1 .Yield surface evolution is self similar, and provided
a triaxial probing over a body of metal powder
confirms this self similarity, metal foam yield flow
behaviour might be thought of as a fairly good
imitation of that of metal powder. This does not
mean that at a micro-scale level the same plastic
flow mechanisms come upon in metal foams and
metal powders.
From relative density 87.5% up until full, theoretical
density: aspect ratio varies in order to recover theVon-Mises yield surface as the material gains full
density. It is by virtue of Eq.(13) that self similarity
breakdown takes place. As full density is neared 0P
tends to infinity and 0Q tends to maxQ . As seen, this
could only be explored by means of curve fitting
functions, for the practical limits of powder
compaction processes prevent compaction densities
close to 100% from being achieved.
Cold compaction of a typical iron powder shows, in
addition, that most of the compaction takes place at
relative densities above 80%, and indicates apractical limit of about 7.2 Kg/m3[10].
For relative densities of 80% and above, Figure 6
shows a point selection fitting a straight line. The
fitting function is given by Eq.(15) thereafter.
-1.20
-1.00
-0.80
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
0. 00 1. 00 2. 00 3. 00 4. 00 5.00 6. 00 7. 00 8. 00 9. 00 10. 00
( )Ln
( )
full
Ln1
1
Figure 6. Linear fitting to compaction data at high
relative densities (>80%)
( ) ( ) 378.011ln177.0ln +
= full
(15)
Eq.(15) is actually a linearization of Eq.(16), which
is the sought evolution equation ( ) .
( ) ( )177.0
11459.1
= full
(16)
Now, the way in which uniaxial yield strength
0 depends on density is also necessary to perform
computations based on the formulations above, andin particular, to make use of Eq.(11). From metal
foam studies, whose specimens are appropriately
subject to uniaxial loading tests, the following
empirical scaling formula is brought in [5,11],
Eq.(17).
( )2
00 3.0
==full
y (17)
Factors 0.3 and 2 come from curve-fitting exercises,
and are used to sort out different metal foam
microstructures; y stands for the yield strength of
the full dense material. Yet, it is noted that this
scaling law embodies the fact that metal foams are
not intended to deform plastically so as to attain full
density: for a relative density of 1.0 yield strength
would only be one third that of the full densematerial. On accounts of that a much simpler
expression, yet more abiding to experimental
observations, is put forward to use, Eq.(18):
=full
y 0 (18)
Microstructure issues of apparent relevance to metal
powders are spared from analysis. Worth to mention
is that a full dense material obtained from powdercompaction may show plastic deformation
mechanisms different from those of wroughtmaterials: contact points between particles are
usually tainted with impurities which might quite
well influence dislocation slip mechanisms.
5 SCALING EQUATIONS FOR ELASTIC
PARAMETERS
In a like manner to uniaxial yield strength
dependence on density, scaling of elastic parameters
is also brought from rather empirical, curve fitting
-
7/26/2019 RLMMArt-09S01N1-p223
9/14
Design of a die for the cold compaction calibration of powdered materials
Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234 229
exercises which, however, have proven useful in
leading to handy, yet accurate models of material
behaviour. So, elastic modulus may be related to
density as shown in Eq.(19) [5,11]:
( )
2
3.0 == fullfullEEE (19)
Once again, 0.3 and 2 account for microstructure
features of some specific metal foam; also,
fullE stands for Young modulus to the full dense
material. Poisson ratio has been said to be 0.3,
typical of many full dense structural metals, and not
bearing a noticeable dependence on density.
Nonetheless, an empirical study performed over
porous structural metals obtained by sintering of
porous compacts provided the following
relationship, employed herein [11]:
( )
==full
37.1exp068.0 (20)
6 RADIAL STRESS COMPUTATIONS
From the discussion following Eq.(11) radial stress
computation comes after two steps: First, define a
set of relative densities spanning over the range of
densities. 1...,,, 210 fullfullfull ; Second,
compute the value of material properties (elastic and
plastic) from evolution equations provided. To eachone of the set of relative densities, compute the
punching pressure P from Eq.(11). This punchingpressure is at equilibrium with elastic deformation
of a porous body of the selected relative density, and
imposes a plastic deformation corresponding to that
relative density as well. Finally, radial stress are
computed following Eq.(10).
From elaborated material data and evolution laws,
corresponding to a quite general class of an iron
alloy porous body, computations are shown in Table
3 (the following values were used:
MPaGPaE yfull 200;200 == ).
Table 3.Material data from compaction of an iron alloy powder [UWS, 2007], and resulting radial stresses.
Relative
densityE y P (MPa)
PFullDense
(MPa)
VonMises
Fraction
Radial
Stress
(MPa)
0.448 1.20E+10 0.13 1.500 8.95E+07 93.41 233.52 0.40 13.410.675 2.73E+10 0.17 1.500 1.35E+08 142.07 252.18 0.56 29.40
0.870 4.54E+10 0.22 1.017 1.74E+08 206.46 281.12 0.73 59.58
1.000 6.00E+10 0.27 0.003 2.00E+08 315.15 315.15 1.00 115.15
In addition to the sought figures for radial stress,
data labelled as P Full Dense and Von Mises
Fraction is provided too. The former refers to the
punching load that should be applied if a full dense
material of corresponding yield strength were to be
taken to its plastic yielding onset. Comparison with
the actual punching pressure P required to incurplastic yield in the porous material provides a sense
of how porosity makes for easier yielding under
applied load, and also the way a full dense, Von-
Mises behaviour is recovered as, indeed, full density
is attained and yield surface aspect ratio reducesto zero. The latter provides a figure of this
recovering process, and is plotted in Figure 7 below.
Point 1 in Figure 7 corresponds to the last value
where a constant aspect ratio of 1.5 is employed.Point 2 corresponds to a relative density of 0.87, and
from this point on the evolution equation Eq.(16) is
employed. There is a noticeable change when
moving on between these ranges, due to the fact of
using models which try to represent data in Table 1
and Table 2. Also the sharp transition from point 3to point 4 is a reflection of how the material model
rushes to recover Von Mises plasticity only at
relative densities close to unity.
Radial stress data are used to estimate die wall
thickness in accordance with the requirements laid
down in the Introduction to this paper. This andother details of the closed die design are exposed
next on.
-
7/26/2019 RLMMArt-09S01N1-p223
10/14
Barrera et al.
230 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
1.10
0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
Relative Density
VonMise
sFraction
1
2
4
3
Figure 7.Approximation to Von Mises-like behaviour at
relative densities close to theoretical.
7 CLOSED DIE DESIGN
Common constitutive equations for mechanicalbehaviour of granular bodies imply a joint evolution
of axial stress, radial stress and porosity within the
specimen, which must be traced out so that the
resulting plots convey the desired material
parameters. This is actually the problem whose
solution is sought when designing a parameter
identification rig. It will be henceforth regarded as
Measuring Instrument, or just MI.
The MI aimed at in this study is as set out below,
Figure 8.
P
Upper punch
Rigid plate
Load cell
Die wall
Powder
Foil strain gage
Figure 8.Scheme of the instrumented closed die.
The upper punch load conveys an axial
stress axial
to the powder body. The die wall is rigid
so that it is a kinematic constraint and a radial stress
radial is developed. These two stresses allow forcomputing stress invariants p and q (or their related
quantities e Eq.(4) and m , Eq.(5)). Upper punch
displacement data permits volumetric strain
computation in turn.
1.1 Specimen size
First considerations come from specimen height to
cross section diameter relationship DH/ : the
larger DH/ , the more inhomogeneous and more
expensive the specimen. However, an DH/ toosmall would give a total punch displacement small
enough so that quite few experimental points could
be taken without interfering with punch travel
device accuracy figure. In addition, failure tests to
the powder body at different relative densities (e.g.
unconfined compression test, Brazilian disc test[12]) might impose a lower bound on DH/ ratioas well. This all adds to the need of making room
for the foil strain gage to capture actual data from
powder compaction. With a bore diameter of
12.5mm as starting point, and a maximum DH/ offour, pressure versus fractional porosity data from
references [11] were used to compute the required
punch displacements to seven figures of fill level
heights (from 30 mm to 116 mm), as set out in
Table 4.
Punch travel for different relative densities anddifferent fill level heights has been computed. As fill
level increases, so does punch travel. For a 30mm
figure, punch travel roughly ranges from 5mm to
10mm. Now it depends on the punch displacement
measuring device the decision whether this range is
large enough to take about 10 experimental points
without concern as to the measuring instrument
accuracy. For the present study a loose powder
cylindrical specimen of 30mm height and 12.5mm
diameter was deemed appropriate. It is worth to
recall, however, that obtaining compacts for failure
tests at different, intermediate densities shall require
fill level heights of at least 30mm.
1.2 Radial stress, hoop strain
radial is computed indirectly from the hoop (or
tangential) strain on die walls outer surface
.
Here the relationship( )radialf = is called
transduction equation, and is obtained as follows,
according to Figure 9.
-
7/26/2019 RLMMArt-09S01N1-p223
11/14
Design of a die for the cold compaction calibration of powdered materials
Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234 231
p i
a
b r
Figure 9. Geometric characterization of the closed die
cross section, for design purposes.
Considering the die wall as a thick walled cylinder,
hoop strain for plane strain condition is given by[13]:
( )[ ]zrvE
+=1
(21)
( ) += rz v (22)
At br= , where hoop strain is to be measured, there
is a free surface, so 0=r and one is left with:
( )[ ]211 vE
=
(23)
It may be found in reference [13] that for a non
rotating, thick walled cylinder:
+
=
2
2
22
2
1r
b
ab
pa i
(24)
Where ip stands for internal pressure. In the present
case this is precisely radial exerted by the loaded
specimen. Enforcing br= and substituting backgives:
( ) ( )
( )
radialradialabE
vaf
==22
22 12 (25)
Table 4.Necessary punch displacements (mm) to achieve densities near to theoretical, for typical metal powders [German,
1994], given several figures of fill level heights.
Material 1 -(Relative Density) P(Mpa) 30 45 60 75 90 100 116
0.35 0 0.00 0.00 0.00 0.00 0.00 0.00 0.00Bronze (Spherical)1
0.04 1000 9.69 14.53 19.38 24.22 29.06 32.29 37.46
0.22 100 0.00 0.00 0.00 0.00 0.00 0.00 0.00Copper (Spherical)2
0.054 520 5.26 7.90 10.53 13.16 15.79 17.55 20.36
0.28 150 0.00 0.00 0.00 0.00 0.00 0.00 0.00Iron (Sponge)2
0.12 610 5.45 8.18 10.91 13.64 16.36 18.18 21.09
0.4 150 0.00 0.00 0.00 0.00 0.00 0.00 0.00Stainless steel 2
0.16 770 8.57 12.86 17.14 21.43 25.71 28.57 33.14
Required punch displacements equal to zero indicate the onset of load application.
-
7/26/2019 RLMMArt-09S01N1-p223
12/14
Barrera et al.
232 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
And this is the transduction equation. Radial
stress within the specimen can be taken from Table
3. The resulting hoop microstrains for a 10mm die
wall thicknesses ( )ab are given in Table 5. Diewall is to be made of tool steel
( 3.0;200 == GPaE ).
Table 5. Radial stresses and hoop microstrains for a
compaction process.
Radial Stress (Mpa) Hoop microstrain
13.41 21.18
32.55 51.42
66.88 105.66
115.15 181.92
The hoop strains shown in Table 5 can be readily
measured by currently available foil strain gages, so
a die wall thickness of 10mm was selected to buildthe compaction die.
1.3 Axial stress, assemblage
The load cell beneath the powder specimen providesdata to asses a correction to upper punch pressure
with due to die wall-powder friction during
compaction. Both upper punch and a bottom plate
are rigid, whereas the load cell body is made of a
linear elastic material whose strain upon load
application is high enough to be captured by means
of an attached foil strain gage. References [3] show
that after total loading, load cell force is of up to
90% that from upper punch. This provides a clue on
design features.
From data in Table 4, upper punch pressure ranges
from about 100MPa the lowest to 1000MPa the
highest. This means a 90-900MPa range on the load
cell. At this point a decision was made in regards of
lower punch cross sectional area: the arrangement of
Figure 10 shows, in addition to the assemblage
description and picture, that this area is the same as
that of the specimen. This ensures that the lower
punch, which acts as a load cell, does not collapse at
high compaction pressures, and also that punchingforce line and bottom punch axis are adequately
aligned.
Metal plates onto the top punch and beneath the
bottom one were added to prevent the pressing
machine plates from denting.
TOP PLATE
UPPER PUNCH
DIE
METAL POWDER
BOTTOM PUNCH
STRAIN GAGE
LEAD PROVISION
HOOP STRAIN
GAGES
Figure 10.Instrumented closed die.
Using a tool steel bottom punch cell would convey a
5000strain under 1000MPa of pressure just overthe surface in contact with the powder. This hints a
limit in top punch applied stress should be
considered. Specifically, and in order to meet
reported punching pressures to attain high
compaction density in a wide variety of metal
powders of industrial interest, upper punch stress is
bounded to 800MPa, So that, at worst, a 3600strain (considering an up to 10% loss of punch
pressure on accounts of die wall-powder friction) is
obtained. A typical foil strain gage holds its
measuring features up until 4000.
1.4 Volumetric strain
Upper punch displacement is directly related tothe volume change of the specimen, and a strain can
be calculated thereof, Figure 11.
Ho
P
Figure 11.Data for volumetric strain calculation
-
7/26/2019 RLMMArt-09S01N1-p223
13/14
Design of a die for the cold compaction calibration of powdered materials
Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234 233
Here oH is the initial, fill level height. From data in
Table 4 may be of up to 10mm for oH of 30mm
(see discussion on specimen size determination).
This gives a volumetric strainV
Vv
= of about
33%. Thus, a logarithmic measure of strain must beconsidered, yielding:
=
==
o
o
o
f
vvH
H
V
V
V
dVd
lnln (26)
Relative density is frequently used to characterize
the material evolution throughout compaction, and
is computed as shown by Eq.(27).
( )( )
=
0
2
4
HD
mR
full
D
Bring m the mass of the specimen.(27)
1.5 Design tolerances
References from powder compaction tooling
designers [14] provided design tolerances and
material specifications (AISI D2 cold work tool
steel; hardness greater than of equal to 62HRC on
contact surfaces), which where also used to check
that elastic deformations under applied loads were
within design limits, and in particular, that radialstraining of the punches inside the die did not
interfere with the punching stress intended to
deform the powder body. Figure 12 shows
construction details.
55mm
MIN12.457mm
MAX12.484mm
MIN12.5mm
MAX12.527mm
75mm
17mm
32.5mm
3mm
15mm
MIN12.457mm
MAX12.484mm
25mm
0.010to 0.01545
Punch
Punch end detail
Bottom Punch
Compaction die
Upper Punch
Figure 12. Construction details of the closed die and
punches.
8 CONCLUDING REMARKS
In regards of the material model employed to solve
the closed die compaction problem in order to get a
figure of radial stress exerted by the material against
the die wall, it is worth remarking that it was a
conjunction of assumed linear elastic, small strain
behaviour, with a porosity dependent yield criterion
and a Von Mises plasticity recovering at high
relative densities. The validity of this model was
assumed on the reported elliptic shape of the yield
surface for metal foams, which were deemed an
appropriate example of porous bodies at low
densities; and that at full densities the material was a
ductile one, so that Von Mises yield surface proves
right.
The model was successfully solved for the close diecompaction case, and in particular, a maximum
value of radial stresses was the expected outcome. It
represents advancement in that, previously, trial and
test prototype building was the manner of finding
out radial stresses at the design stage of
instrumented dies.
Concerning the instrumented die construction three
hoop strain gages were attached to the outer surface
of the die. They are aimed at capturing a radial
stress variation along the specimen length on
accounts of sliding die wall friction. The setup was
used to study this problematic [15], whose resultsare expected to be published in the near future
Accuracy and precision of measurements can be
computed from basic standard data from strain
gages along with computations from Eq.(25) and
data in Figure 12 ( 3.0;200 == GPaE are takenfrom general references so their scatter is considered
zero); and also from pressing machine load cell
accuracy and cross section area of top punch,
bearing in mind its tolerances from Figure 12 as
well. Further refinements are aimed at specifying
uncertainty figures to alignment/misalignment of
hoop strains when attached onto the die surface.
9 REFERENCES
[1] Standard ASTM B331-85: Compressibility of
Metal Powders in Uniaxial Compaction, Vol.
02.05American Society for Testing and
Materials, 1990.
[2] U.W.S.(University of Wales Swansea), Civil andComputational Engineering Centre. MERLIN
-
7/26/2019 RLMMArt-09S01N1-p223
14/14
Barrera et al.
234 Rev. LatinAm. Metal. Mater.2009; S1(1): 223-234
powder compaction software: users manual.
2007.
[3] Zavaliangos, A., Cunningham, J., Sinka, I.C. J.
of Pharm. Sci. 2004; 93 (8).
[4] Budynas, R. Advanced Strength and Applied
Stress Analysis. McGraw-Hill, 1999.
[5] Ashby, M.F. et al. Metal Foams: a Design
Guide. Butterworht-Heinemann, 2000.
[6] Heckel, R.W. Trans. Metall. Soc. AIME. 1961;
(221): 671.
[7] Deshpande, V.S., Fleck, N.A. J. Mech. Phys.
Solids. 1999; (48): 1253-1283.
[8] Green, R.J. International Journal of Mechanical
Sciences. 1972; (14): 215-224.
[9] Terzaghi, K. Soil Mechanics in Engineering
Practice. 2nd Edition. John Wiley & Sons,
1967.[10] Hoganas. Iron Powder Handbook. 1957; 1.
[11] German, R. Powder Metallurgy Science, 2nd
Edition. MPIF, Princeton, NJ., 1994.
[12] Stagg, R.G. Rock Mechanics in Engineering
Practice. John Wiley & Sons, 1968.
[13] Timoshenko, S. Theory of Elasticity, 3rdEdition. McGraw-Hill, 1970.
[14] Kunkel, R. Manufacturing Engineering &
Management. 1970; 240-244.
[15] Barrera, M. Construction of a True Compaction
Curve: Critical Review of the Closed DieCompaction Test for Powdered Materials.
PhD Thesis Cali (Colombia), Materials
Engineering School, Universidad del Valle,
2008.